胃癌术后教科书式结局影响因素分析及预测模型构建

Analysis of Influencing Factors of Textbook Outcome after Gastric Cancer Surgery and Construction of Predictive Model

摘要: 目的：探讨胃癌根治术后患者获得教科书式结局的影响因素，并构建预测模型。方法：回顾性收集2018年9月至2023年4月安徽医科大学第一附属医院胃肠外科行胃癌根治术的394例患者临床资料，随机分为训练集(315例)和验证集(79例)。通过单因素及多因素Logistic回归分析筛选TO的独立影响因素，并构建Nomogram预测模型，采用受试者工作特征曲线评估模型性能。结果：多因素分析显示，白蛋白 < 35 g/L (OR = 0.240, P < 0.001)、CEA升高(OR = 0.103, P < 0.001)、CA125升高(OR = 0.171, P = 0.012)及肿瘤直径 > 3 cm (OR = 0.413, P = 0.016)为TO的独立危险因素，而肿瘤分期升高为独立保护因素(OR = 1.459, P = 0.002)，分析可知该结果系统计学导致的人为假象。基于上述因素构建的Nomogram模型在训练集和验证集中的AUC值分别为0.831和0.812。结论：术前白蛋白水平、CEA、CA125、肿瘤直径及肿瘤分期是胃癌术后TO的独立影响因素，基于此构建的预测模型具有良好的预测效能。

Abstract: Objective: To investigate the factors influencing textbook outcomes of patients with gastric cancer after radical gastrectomy and to construct predictive models. Methods: The clinical data of 394 patients undergoing radical gastrectomy in the Department of Gastrointestinal Surgery of the First Affiliated Hospital of Anhui Medical University from September 2018 to April 2023 were retrospectively collected and randomly divided into training set (315 cases) and validation set (79 cases). The independent influencing factors of textbook outcome were screened by univariate and multivariate Logistic regression analysis, and Nomogram prediction model was constructed. The performance of the model was evaluated by subject operating characteristic curve. Results: Multivariate analysis showed that albumin < 35 g/L (OR = 0.240, P < 0.001), elevated CEA (OR = 0.103, P < 0.001), elevated CA125 (OR = 0.171, P = 0.012) and tumor diameter > 3 cm (OR = 0.413, P = 0.016) were independent risk factors for TO, while elevated tumor stage was an independent protective factor (OR = 1.459, P = 0.002), our analysis reveals that this finding is likely a statistical artifact.. The AUC values of Nomogram model constructed based on above factors are 0.831 and 0.812 in training set and validation set respectively. Conclusion: Preoperative albumin level, CEA, CA125, tumor diameter and tumor stage are independent predictors of TO after gastric cancer surgery. The predictive model based on these predictors has good predictive power.